This invention relates generally to the field of control systems and more particularly to an active noise control system and method for on-line feedback path modeling.
Active noise control systems are concerned with the reduction of any type of undesirable disturbance or noise signal provided by a noise source through an environment, whether it is borne by electrical, acoustic, vibration, or any other kind of noise media. Since the noise source and environment are often time-varying, the noise signal will often be non-stationary with respect to frequency content, amplitude, and velocity. Active noise control systems control noise by introducing a canceling xe2x80x9canti-noisexe2x80x9d signal into the system environment or media through an appropriate secondary source. The anti-noise signal is ideally of equal amplitude and 180 degrees out of phase with the noise signal. Consequently, the combination of the anti-noise signal with the noise signal at an acoustical summing junction results in the cancellation or attenuation of both signals and hence a reduction in noise.
In order to produce a high degree of noise signal attenuation, the amplitude and phase of both the noise and anti-noise signals must match closely as described above. Generally, this is accomplished by an active noise control system using an active noise control system controller that performs digital signal processing using one or more adaptive algorithms for adaptive filtering. The adaptive filtering, and more specifically the adaptive algorithms, track all of the changes in the noise signal and the environment in real-time by minimizing an error signal and continuously tracking time variations of the environment. The adaptive filtering may use any of a variety of known and available adaptive algorithms, such as the least-mean-square (xe2x80x9cLMSxe2x80x9d) algorithm, to establish the taps or coefficients of an associated adaptive filter that models the noise source and environment to reduce or minimize the error or residual signal.
Active noise control systems, as compared to passive noise control systems, provide potential benefits such as reduced size, weight, volume, and cost in addition to improvements in noise attenuation. Active noise control is an effective way to attenuate noise that is often difficult and expensive to control using passive means and has application to a wide variety of problems in manufacturing, industrial operations, and consumer products.
Active noise control systems may generally be divided into feedforward active noise control systems and feedback active noise control systems. The present invention will be illustrated as applied to a feedforward active noise control system and thus the present invention will be described in this context.
A feedforward active noise control system generally includes a reference sensor for sensing a noise signal from a noise source and generating a corresponding primary signal in response; an active noise control system controller for generating a secondary signal; a secondary source, located downstream from the reference sensor, for receiving the secondary signal and generating an anti-noise signal to cancel or attenuate the noise signal; and an error sensor for detecting a residual signal and generating a corresponding error signal in response. The residual signal is equivalent to the difference between the noise signal and the anti-noise signal as provided to the error signal through a primary environment. The active noise control system controller receives the primary signal and the error signal and generates the secondary signal in response.
The active noise control system controller is implemented using a digital signal processor and performs digital signal processing using a specific adaptive algorithm, depending on the type of cancellation scheme employed, for adaptive filtering. Also, the reference sensor, the secondary source, and the error sensor may include interface circuitry for interfacing with the active noise control system controller. The interface circuitry may include analog-to-digital converters, digital-to-analog converters, analog filters such as low pass filters and automatic gain control amplifiers so that signals can be exchanged in the correct domain, i.e., either the digital or analog domain. The interface circuitry may be provided separately.
Feedforward active noise control systems include a primary path that has a transfer function that may be denoted as P(z). The primary path may be defined as the environment from the reference sensor to the error sensor. Feedforward active noise control systems also include a secondary path and a feedback path. The secondary path has a transfer function that may be denoted as S(z). The secondary path may be defined as the environment from the output of the active noise control system controller to the output of the error sensor. This may include interface circuitry such as a digital-to-analog converter, an analog filter, a power amplifier, a loud speaker, an error microphone, and other devices. The feedback path also has a transfer function and may be denoted by F(z). The feedback path may be defined as the environment from the output of the active noise control system controller to the output of the reference sensor. The active noise control system controller, using a digital signal processor, may include an adaptive filter, that is normally denoted by W(z), that attempts to adaptively model the primary path. The objective of the adaptive filter W(z) is to minimize the residual signal or error signal. The adaptive filtering performed by adaptive filter W(z) may be performed either on-line or off-line.
Feedforward active noise control systems suffer from a serious drawback that often harms overall system performance. Whenever the secondary source generates an anti-noise signal to cancel the noise signal, a portion of the anti-noise signal radiates upstream to the reference sensor where it is received along with the noise signal. The path that the anti-noise signal takes when traveling from the secondary source to the reference sensor is the feedback path. The feedback path, once again, may be defined as the media environment from the output of the active noise control system controller to the output of the reference sensor. The portion of the anti-noise signal flowing to the reference sensor along the feedback path is part of a feedback signal that travels through the feedback path. As a consequence of the feedback signal being received at the reference sensor, an incorrect primary signal is provided to the active noise control system controller by the reference sensor and, hence, overall system performance is harmed. If the feedback signal is in phase with the noise signal, the reference sensor will generate a primary signal that is too large. If the feedback signal is out of phase with the noise signal, the reference senor will also generate a signal that is incorrect. In any event, the feedback signal is undesirable and harms overall performance. The feedback signal may also allow the introduction of poles into the response of the system transfer function which results in potential instability if the gain of the feedback loop becomes large.
In certain applications, overall system performance is significantly degraded if the effects of the feedback path are not modeled and neutralized. The modeling of the feedback path and neutralization of the feedback signal becomes especially critical to overall active noise control system performance in applications in which the secondary source is in close proximity or in close communication with the reference sensor. Such systems would include, for example, appliances such as refrigerators and window air conditioner units in which the air ducts are relatively short. In such applications, the secondary source must be located close to the reference sensor by necessity and hence the feedback signal and its adverse effects will be greater.
The feedback path problem has been recognized in the past and several solutions have been proposed with limited success. A first set of proposed solutions has focused on the use, type, and placement of the reference sensors and the secondary sources, while a second set of proposed solutions has focused on signal processing techniques. The first set of proposed solutions involves the use and placement of directional reference sensors and secondary sources to limit or minimize the feedback signal. These proposed solutions add additional expense and complexity to the system and decrease overall reliability while making it difficult, if not impossible, to obtain good directivity over a broad range of frequencies.
The second set of proposed solutions has focused on signal processing techniques and has achieved limited success. The proposed solutions involving signal processing techniques may be generally separated into off-line modeling techniques and on-line modeling techniques. Both off-line modeling and on-line modeling are system identification techniques in which a signal is provided to the system and the resulting signal is analyzed to construct a model of the unknown system. This is accomplished by exciting an unknown path or environment with the known signal and then measuring or analyzing the resulting signal that is provided in response.
Off-line feedback path modeling techniques involve providing a known signal in the absence of the noise signal cancellation that is normally provided by the active noise control system. An adaptive algorithm is used to calculate the coefficients or taps of an adaptive filter to minimize the effects of the feedback path. Once the coefficients or taps are established off-line, during actual active noise control system operation, the taps or coefficients are fixed in a digital filter and are not changed during actual operation. Although off-line feedback path modeling techniques are adequate in certain situations, off-line modeling may not provide adequate performance when used in a system in which parameters are frequently changing. For example, parameters such as temperature and signal flow rate may frequently change resulting in an inaccurate feedback path model because of the changes.
Another problem with off-line feedback path modeling is that the noise signal must be eliminated or stopped for the off-line feedback path modeling to correctly model the unknown environment. This is often not practical in many real-world systems. For example, a power transformer that is energized and used to provide power to customers cannot be easily taken out of service so that off-line modeling may take place. In a system that changes frequently, it may be necessary to routinely perform off-line feedback path modeling so that the feedback path remains accurately modeled. In the event that a noise source cannot be shut off, off-line modeling may proceed if the known signal or modeling signal is provided at a very high amplitude for an extended period of time. In spite of this, the off-line model may still be inaccurate.
On-line feedback path modeling refers to the modeling of the feedback path while the noise signal is being provided to the unknown environment and the active noise control system is operating to cancel the noise signal. Ideally, on-line feedback modeling allows for any changes in the plant environment to be modeled while the active noise control system is operating and thus avoiding the problems encountered with off-line feedback path modeling when the environment or plant changes due to such things as temperature and flow changes. Unfortunately, prior attempts at providing on-line feedback path modeling have proven unsatisfactory and have failed to provide an on-line model of the feedback path.
One such technique focused on providing an adaptive neutralization filter in parallel with the feedback path. The adaptive neutralization filter approach, such as that described in U.S. Pat. No. 4,473,906 entitled xe2x80x9cActive Acoustic Attenuator,xe2x80x9d may only effectively operate in an off-line feedback path modeling mode because of the fact that the adaptive neutralization filter will attempt to adapt even when the noise signal and the anti-noise signal are perfectly canceled. The feedback neutralization technique attempts to model the feedback path in such a way as to remove all portions of the primary signal that are correlated with the output of the adaptive filter, which, ideally, results in a system that appears to be without feedback. Since the primary noise signal is highly correlated with the anti-noise signal, the adaptive feedback neutralization filter will continue adapt even when the feedback signal is perfectly canceled. As a consequence, the adaptation of the feedback neutralization filter must be deactivated when the system is on-line. Also, when the noise signal contains narrowband frequency components, the adaptive feedback neutralization filter may fail to properly converge when attempting to adapt on-line.
Another proposed on-line feedback path modeling solution involves the use of an infinite-impulse response (xe2x80x9cIIRxe2x80x9d) filter to compensate for the feedback signal. This approach has achieved only limited success. For example, in U.S. Pat. No. 4,677,677 entitled xe2x80x9cActive Sound Attenuation System with On-Line Adaptive Feedback Cancellation,xe2x80x9d an adaptive IIR filter structure was proposed for use in an active noise control system. In this approach, the feedback path is considered part of the overall plant model but does not truly model the feedback path. This approach suffers several disadvantages which are inherent in adaptive IIR filters. For example, IIR filters are not unconditionally stable because of the possibility that some poles of the IIR filter will move outside of the unit circle during the adaptive process, resulting in instability. Also, due to the presence of local minima the adaptation may converge at one of the local minima. Furthermore, adaptive algorithms used with IIR filters often have a relatively slow convergence rate in comparison with that of FIR filters.
Other proposed on-line feedback path modeling solutions involve the use of a modeling signal that must be provided at a very high amplitude so that it may be distinguished from the noise signal. This solution introduce additional noise into the system that adversely affects overall active noise control system operation and performance.
From the foregoing it may be appreciated that a need has arisen for an active noise control system and method for on-line feedback path modeling that eliminate or reduce the problems described above. In accordance with the present invention, an active noise control system and method for on-line feedback path modeling are provided that provide a signal processing solution to the feedback signal problem by providing on-line modeling of the feedback path and neutralizing its effects so that an active noise control system will operate more efficiently and accurately. This is accomplished even when the feedback path is changing. The present invention attenuates both broadband noise signals and narrowband noise signals.
According to an embodiment of the present invention, an active noise control system is provided for generating an anti-noise signal to attenuate a noise signal provided through a media of a primary path. The active noise control system performs on-line feedback path modeling and feedback path neutralization. The active noise control system includes a reference sensor, a secondary source, an error sensor, and an active noise control system controller. The reference sensor receives the noise signal and a feedback signal and generates a primary signal in response. The secondary source receives a secondary signal and generates a corresponding anti-noise signal in response. The anti-noise signal is provided to the media to attenuate the noise signal. The error sensor receives a residual signal that is a combination of the noise signal and the anti-noise signal as received at the error sensor. The error sensor generates an error signal in response to receiving the residual signal. The active noise control system controller receives the primary signal and the error signal and generates the secondary signal while performing on-line feedback path modeling.
The present invention provides various technical advantages. A technical advantage of the present invention includes the ability to accurately perform on-line feedback path modeling to improve overall active noise control system performance. Another technical advantage of the present invention includes the ability to implement the present invention using existing digital signal processing techniques and algorithms. Yet another technical advantage of the present invention includes increased active noise control system stability due to the elimination of the feedback path effects. Still another technical advantage of the present invention includes the ability to cancel or attenuate both broadband and narrowband noise signals. Other technical advantages are readily apparent to one skilled in the art from the following FIGUREs, description, and claims.